Sorry Traci -- I'm usually good about including this: login pub password download
On 10/08/2009 03:16 PM, Sandoval, Traci I wrote: > Donna, the link appears to be password protected, thank you Traci > > http://brainmap.wustl.edu/pub/donna/FREESURFER/SCRIPTS/2009_09/ > > -----Original Message----- > From: [email protected] > [mailto:[email protected]] On Behalf Of Donna Dierker > Sent: Thursday, October 08, 2009 1:42 PM > To: Caret, SureFit, and SuMS software users > Subject: Re: [caret-users] do spherical registration using freesurfer > generated surfaces > > Hi Traci, > > No, I don't. *Someone* -- YOU KNOW WHO YOU ARE -- has me working on > similar scripts, but different enough to be both a fair amount of work > and not as applicable to what I had in mind. I hope to wrap that up this > week (but we'll see). > > Meanwhile, I posted interim versions of what I do have: > > http://brainmap.wustl.edu/pub/donna/FREESURFER/SCRIPTS/2009_09/ > > freesurfer2caret.Ver2a.preborder.mni2wustl.sh - imports Freesurfer to > Caret and creates inputs for next script > draw_core6_landmarks.tim.sh - draws borders > border_variability.sh - generates quantitative measures of landmark quality > generate_scene.check.sh - generates captures, scenes of borders > [manual step: Layers: Borders: Border Update] > postborder.sh - smooths medial wall, morphs and aligns sphere, registers > > If you have more than a handful of subjects, it will probably be worth > getting these to work. > > Realistically, I hope to have something applicable to more typical > Freesurfer input by 10/31. > > The script above had more steps like this than the typical stream would > have, due to strangeness of the input data and inadequate talairach.xfm > (not imgreg's fault): > > caret_command -surface-apply-transformation-matrix > $CASE.$HEM_FLAG.MIDTHICKNESS.coord $CASE.$HEM_FLAG.topo > $CASE.$HEM_FLAG.MIDTHICKNESS_orig.coord -matrix $MATRIX > > Donna > > > On 10/08/2009 01:15 PM, Sandoval, Traci I wrote: > >> Hi Donna, do you have these cool caret scripts ready ie... >> >> Next week, I'll adapt these import scripts to a dataset more like what yours >> is likely to be. When I am done, I can post the resulting scripts. >> >> >>> Can you ping me in two weeks? >>> >>> Donna >>> >>> >> Thanks for all your work and help. Traci >> >> Traci Sandoval >> Research Assistant >> NeuroPsychometric Research Lab >> School of Behavioral and Brain Sciences - University of Texas at Dallas >> Center for BrainHealth - University of Texas at Dallas >> Department of Psychiatry - University of Texas Southwestern Medical >> www.utdallas.edu/research/nprlab >> 972-883-3270 >> >> >> >> -----Original Message----- >> From: [email protected] >> [mailto:[email protected]] On Behalf Of Donna Dierker >> Sent: Friday, September 18, 2009 11:20 AM >> To: Caret, SureFit, and SuMS software users >> Subject: Re: [caret-users] do spherical registration using freesurfer >> generated surfaces >> >> Traci, >> >> Yes, one lab here at Wash U is doing that. A few comments: >> >> * Since you don't have a flat map, the reference distance for the >> posterior sylvian starting point changes from 9mm posterior to the >> SF/SF2 intersection (relative to the flat map) to 12mm posterior to the >> SF/SF2 intersection on the spherical map. >> >> * It's tougher to discern the full sylvian trajectory on the spherical >> map, as well as center the sylvian on the main window, such that you can >> see the whole length. Fortunately, the absolutely fun-to-use Layers: >> Borders: Border Update makes this easy to fix, if you can't draw the >> whole length in one go. (If your older Caret version doesn't have this >> feature, you really MUST upgrade.) >> >> * You have to draw the medial wall dorsal and ventral in separate >> segments, and you have to pay attention to their orientation (dorsal >> starts anteriorly, ventral posteriorly). Calcarine starts posterior. >> Flattening took care of this for you. >> >> * The lack of a compressed medial wall view makes it harder to draw the >> medial wall ventral. >> >> Donna >> >> On 09/18/2009 10:19 AM, Sandoval, Traci I wrote: >> >> >>> Donna, That’s sounds great, I have scripts that take my freesurfer brains >>> all the way to flattening in caret, but I flatten and draw borders >>> manually. I noticed you mentioned you guys are doing the auto-landmarks on >>> the spherical surface, and not flattening at all. Can I do that manually as >>> well? And skip flattening? Thanks Traci >>> >>> -----Original Message----- >>> From: [email protected] >>> [mailto:[email protected]] On Behalf Of Donna Dierker >>> Sent: Friday, September 18, 2009 9:26 AM >>> To: Caret, SureFit, and SuMS software users >>> Subject: Re: [caret-users] do spherical registration using freesurfer >>> generated surfaces >>> >>> Traci, >>> >>> I do have some scripts I could make available now, but in two-four >>> weeks, I'll have more general purpose ones. The most complete set were >>> actually developed for a data set segmented in Caret. But they include >>> the auto-landmarks, border variability (QA), generating scenes/captures >>> showing the borders, and running registration via command line. There >>> also are several versions of a "preborder" script that covers importing >>> from Freesurfer, but most of these used anatomical data that was in >>> 711-2* space before feeding to Freesurfer. Your data is unlikely to fit >>> that category. >>> >>> Next week, I'll adapt these import scripts to a dataset more like what >>> yours is likely to be. When I am done, I can post the resulting scripts. >>> >>> Can you ping me in two weeks? >>> >>> Donna >>> >>> On 09/17/2009 04:07 PM, Sandoval, Traci I wrote: >>> >>> >>> >>>> Donna, I would love to use these scripts that aren't publicized yet. How >>>> can I attain said scripts?!! Thank you for all your work and questions >>>> answered. Traci >>>> >>>> >>>> Traci Sandoval >>>> Research Assistant >>>> NeuroPsychometric Research Lab >>>> School of Behavioral and Brain Sciences - University of Texas at Dallas >>>> Center for BrainHealth - University of Texas at Dallas >>>> Department of Psychiatry - University of Texas Southwestern Medical >>>> www.utdallas.edu/research/nprlab >>>> 972-883-3270 >>>> >>>> >>>> >>>> -----Original Message----- >>>> From: [email protected] >>>> [mailto:[email protected]] On Behalf Of Donna Dierker >>>> Sent: Thursday, September 17, 2009 8:44 AM >>>> To: Caret, SureFit, and SuMS software users >>>> Subject: Re: [caret-users] do spherical registration using freesurfer >>>> generated surfaces >>>> >>>> Hi Jidan, >>>> >>>> See inline replies below. >>>> >>>> Donna >>>> >>>> On 09/17/2009 02:06 AM, z丹丹 wrote: >>>> >>>> >>>> >>>> >>>>> Hi, >>>>> >>>>> I already have fiducial surfaces generated by Freesurfer, which are >>>>> all in the MNI space. I want to do spherical registration to an atlas >>>>> spherical surface. The steps i need to do are: >>>>> >>>>> 1, generate the ellipsoid, spherical surfaces, and flat maps. >>>>> 2, create the 6 landmarks on both individual and atlas surface. >>>>> 3, do spherical mapping. >>>>> >>>>> Are the procedure right? Thanks! >>>>> >>>>> >>>>> >>>>> >>>> Funny you should ask. We (the Van Essen Lab) are using Freesurfer for no >>>> less than five projects, and we have shell scripts that streamline much >>>> of the import, border drawing, QA, and registration. There is a feature >>>> we call "auto-landmarks" that draws the borders reasonably well. Many >>>> borders need touch-ups, but a single rater fixed 72 hemispheres' borders >>>> in 1.5 days. If you have only a handful of subjects, it's not clear >>>> whether it would be worth it for you to use our methods. But if you have >>>> dozens, then they would save you a lot of time. >>>> >>>> We haven't broadly publicized these features/scripts, because they're >>>> still in the final stages of development and testing. But they have been >>>> working well for us. >>>> >>>> Even if you don't use auto landmarks or these scripts, one question >>>> stands out in your steps above: Do you need a flat map? No, not really. >>>> It certainly is easier to draw landmarks on the flat map than the >>>> sphere; however, I'm not sure it's worth the hassle. Then again, if >>>> you're not using the scripts, it might be the shortest path to >>>> registration. >>>> >>>> >>>> >>>> >>>>> In fact before I move to the 2nd step, I was stuck in the flat map >>>>> generation. I met two problems: >>>>> 1, As my surface is in MNI space, the surface orientation is not right >>>>> in the CARET software. I found that the anterior and posterior are >>>>> reversed with superior and inferior. Is there any method to roate it >>>>> into the right orientation in CARET? >>>>> >>>>> >>>>> >>>>> >>>> Yes, but this is not consistent with my experience of MNI space nor >>>> Freesurfer surface orientation. While MNI volumes are sometimes stored >>>> using left-handed coordinate systems (what some call radiological >>>> orientation), the coordinates are always reported with -x being left, +x >>>> being right; -y being posterior, +y being anterior; -z being inferior, >>>> +z being superior. I've never seen a Freesurfer surface with the Y and Z >>>> axes flipped -- not even ones written in MNI space. >>>> >>>> It's easy to flip a surface from LAS to LPI >>>> (http://brainmap.wustl.edu/OLD/SureFit/orient.html) like this: >>>> >>>> caret_command -surface-apply-transformation-matrix $COORD $TOPO >>>> $FLIPPED_COORD -matrix "1 0 0 0 0 -1 0 0 0 0 -1 0 0 0 0 1" >>>> >>>> But I really would not brush this off. I would get to the bottom of why >>>> the supposedly MNI surface is not in the expected orientation. The >>>> evidence suggests it is not in MNI space, as asserted. >>>> >>>> >>>> >>>> >>>>> 2, The other problem is , I found that the origin of my surface is not >>>>> AC. So when I flatten the surface, it can't find the correct medial >>>>> wall to cut. Do you think there is a way to solve this too? >>>>> >>>>> >>>>> >>>>> >>>> This is more evidence that the surface is not really in MNI space. How >>>> did it get transformed to MNI? >>>> >>>> Note that the MNI origin is not exactly the AC; see >>>> http://imaging.mrc-cbu.cam.ac.uk/imaging/MniTalairach for details. But >>>> it is certainly close enough to the AC that the compressed medial wall >>>> view should be centered over the medial wall. The fact that it is not >>>> indicates a problem with your transformation to MNI space. >>>> >>>> >>>> >>>> >>>>> Thanks a lot! >>>>> >>>>> Jidan >>>>> >>>>> >>>>> _______________________________________________ caret-users mailing list [email protected] http://brainvis.wustl.edu/mailman/listinfo/caret-users
